A Fast Non-parametric Approach for Local Causal Structure Learning
This provides a practical tool for researchers in fields like computational biology who need causal inference without parametric assumptions, though it is incremental as it builds on existing FOCI methods.
The authors tackled the problem of causal structure learning without strong assumptions on functional forms or noise, developing DAG-FOCI, a fast non-parametric algorithm that outputs the parents of a response variable, with theoretical guarantees for acyclic cases and conservative error control for identifiable settings.
We study the problem of causal structure learning with essentially no assumptions on the functional relationships and noise. We develop DAG-FOCI, a computationally fast algorithm for this setting that is based on the FOCI variable selection algorithm in~\cite{azadkia2021simple}. DAG-FOCI outputs the set of parents of a response variable of interest. We provide theoretical guarantees of our procedure when the underlying graph does not contain any (undirected) cycle containing the response variable of interest. Furthermore, in the absence of this assumption, we give a conservative guarantee against false positive causal claims when the set of parents is identifiable. We demonstrate the applicability of DAG-FOCI on simulated as well as a real dataset from computational biology~\cite{sachs2005causal}.